JUNIOR CREATIVE STRATEGIST
Hello, I'm Michael. For the last couple of years I've been building consumer brands on my own, with no team and not much of a budget. The first ended when the law changed. The second I closed myself. What I want next is to do it somewhere I'm not the most experienced person in the room.
Skip to portfolioA nicotine-free vape brand built on one idea: people quitting smoking don't just miss the nicotine, they miss having something to do with their hands. I got it off the ground on organic content alone. It was starting to work when the 2025 disposable ban closed the whole category.
I started Myst at the end of 2023, on a hunch about what people actually miss when they quit. Not the nicotine so much as the ritual: the thing in the hand. So the product kept the habit and dropped everything harmful. I had almost no money behind it, which meant I had to rely on statics and low-prod clips filmed in the house, and a lot of trial and error.
It worked well enough that I was about to commit to a large stock order. Then the 2025 single-use vape ban took the category with it, and I was forced to close the doors.
I had no system back then. I ran on instinct and copied whatever looked like it was working, and a fair bit of it did. What I couldn't do was tell you why, or repeat it on purpose. It gave me a feel for the difference between good and bad creative before I had any language for it. I still look back on it fondly.
A men's skincare brand built around beef tallow, which I didn't believe in either until it sorted my own skin out. The cost of goods was tight from the day I started, and I backed my creative to out-run it. It got a lot closer than I expected. Not close enough.
Rendered started from a more personal place. I'd spent years fighting my skin with products that never quite worked, and I didn't expect a rendered animal fat to do much either. Then it turned out to be the thing that did, so I thought why not build a brand around it.
The real change was leading with psychology. Before anything gets made I go and find the customer: the forums they read, the content they watch, the words they use when nobody is selling to them. What I'm after is the reason they actually buy, which usually sits a rung underneath the reason they'd give you. Then I write to that, and I set the testing up so each result feeds the next instead of starting over. The aim is to know why something worked, so I can do it again on purpose.
I leaned hard on systems this time. Every persona, every angle and every test result lived in one place and fed the next batch, so nothing got worked out twice. Eventually the tooling I wanted didn't exist, so I built it myself.
I closed it in July 2026, and the reason was the cost of goods. Landed cost against what a single tub could sell for left very little room to buy a customer with, and I knew that when I started. The bet I made was that creative could out-run it: that if the ads were good enough, I could pull acquisition down far enough for thin margins to survive.
It nearly worked. The last batch produced the best ad the account ever ran, a 4.9% CTR on cold traffic, and it did drag acquisition costs down. Just not far enough: nCAC landed around £24 against a break-even near £20, and closing that last stretch needed margin I didn't have.
What I took from it is that good creative can rescue a lot of things. It can't rescue arithmetic that was decided by a supplier.
The lesson from Rendered was that a product has to work on paper before anything else about it can work. So the next one is being made in house rather than sourced, in a category I've wanted to build in for years. More on that when there's something to show.
I'm keeping the category to myself for now. What I will say is that it's something I use, in a space I've spent years around rather than one I found in a trends tool, and that the version I want doesn't really exist yet.
Alongside it I'm taking on freelance creative strategy work, and I'm open to a role. Both for the same reason: I want reps on brands that aren't my own, and I'd like to watch how people who have done this longer than me actually operate.
It runs on top of Claude, and it holds the whole brand: the ad pipeline, the numbers, the research, and everything I've read that was worth keeping. The point was to stop losing things, so no insight sits in a chat log I never open again. Keep scrolling and I'll walk you round.
This is Atom. It runs on top of Claude, and it's where everything lives: the daily loop, the ad account, the research, and the creative warehouse.
Every ad gets a card on this board. The same batch flips by status, funnel stage or awareness stage.
Behind each card sits the full record: the creative, its routing and format, and the exact Meta copy. Results from the Meta API land against that same record, so I can track what happened, work out why, and build the next batch on data instead of guesswork.
The Control Room is my daily read: Shopify and Meta pulled into one screen, revenue against spend, the funnel from session to purchase.
This is Atom. It runs on top of Claude, and it's where everything lives: the daily loop, the ad account, the research, and the creative warehouse.
Every ad gets a card on this board. The same batch flips by status, funnel stage or awareness stage.
Behind each card sits the full record: the creative, its routing and format, and the exact Meta copy. Meta API results land against that same record.
Shopify and Meta pulled into one screen: revenue against spend, and the funnel from session to purchase.
Six pieces across both brands, each named for the format it was built to. Tap any of them for the full-size image and the thinking behind it.
It starts with the customer's psychology, long before anything gets ideated. This is how I worked on Rendered, and how I'd work on anything else.
Before I make anything I go and find the customer in their own words. Reviews, Reddit threads, YouTube comments, competitor ads. What I'm after is the reason they actually buy, which normally sits a rung or two under the reason they'd give you.
The raw research becomes three documents per persona: a deep profile, a dossier of the moments the problem actually bites, and a bank of angles tagged so I can pull from it later. Built once, then kept current by the systems rather than by hand. After that, every ad is pulling from language the customer used instead of language I invented.
Before anything gets concepted, I spread the batch across a signal map: awareness stage, funnel stage, self-concept, valence, language intensity. Each ad has to hand Meta a different signal from the one next to it.
Every test gets logged with a hypothesis for why it won or lost, written against what I expected going in. Most of this runs in Atom now. The pipeline links to the Meta API, so the ad, its image, its headline and its numbers sit on one record and I can run my analysis prompts across the lot. The point is that the reasoning carries into the next batch instead of getting lost between them.
Outside the brands I compete in strongman at national level, in the 80kg class. Training gets the same treatment as an ad account: a log, a hypothesis, and a lot of patience with slow feedback.
Most of what I know about marketing came out of old direct response. Schwartz, the Halbert letters, Collier. These days it's people like Sarah Levinger and the Soar With Us lot, and whatever the last test taught me. I've never worked inside a marketing team, and that's the gap I'm trying to close.
What I want next is to do this with more behind me than I've had on my own. I take on freelance creative strategy work, and I'm open to a role at a DTC brand or an agency, somewhere I can learn from people who have been doing this stuff a lot longer than I have.
And if you're a founder who landed here from a message I sent, that's the other half of it. I make spec ads for brands I like, free, and the only thing I ask back is permission to put them on this page.